While an IIT-Kanpur study has predicted a fourth wave of COVID-19 in the country from June to October, three experts in Karnataka that the State’s Technical Advisory Committee (TAC) consulted on this possibility are sceptical of the study.
The paper by researchers from IIT-Kanpur forecasts that the fourth wave of COVID-19 in India will arrive after 936 days from the reporting of the first case, which is January 30, 2020. Therefore, the fourth wave starts on June 22, 2022, reaching its peak on August 23, 2022, and ends on October 24, 2022, according to the paper.
As the same team from IIT-Kanpur had precisely predicted the third wave of Omicron variant in January–February 2022, the State’s TAC had sought an expert review on the paper by Rajesh Sundaresan from IISc, Siva Athreya from Indian Statistical Institute, and TAC member Giridhar R. Babu from the Public Health Foundation of India.
According to the expert reviews, the IIT-Kanpur’s paper suffers from several limitations. “As the predictions are solely based on statistical models, they should be read with caution,” the experts have opined.
The TAC deliberated upon the expert reviews at its 161st meeting held on March 16 and recommended the State to adopt a cautious approach irrespective of the paper’s limitations.
Prof. Sundaresan has said the paper uses Zimbabwe as the training data since “Zimbabwe and India have the maximum visible similarities in the shape of the COVID waves”.
“But Zimbabwe and India have very different vaccinated populations. Roughly, 30% have taken at least one dose and 23% two doses in Zimbabwe. The situation in India is vastly different (where vaccination is 70% first dose and 59% second dose). These higher vaccination numbers may significantly delay a fourth wave and may result in a different wave width,” he said.
“The Zimbabwe wave in November was likely due to Omicron. To say Zimbabwe’s Omicron wave width will apply to India’s fourth wave (which will possibly be due to a new variant) is without basis. A new dominant variant has to overcome Omicron’s prevalence and possible hybrid immunity (due to Delta/Omicron infections and vaccinations),” Prof. Sundaresan said.
Pointing out that the prediction suffers from several limitations, Dr. Babu said the major drawbacks of the Gaussian model (used in the prediction) are that there is a tendency of overfitting the data and high sensitivity to initial conditions.
“The natural progression of the pandemic depends on several factors. Among them, the actions taken during the initial days of the outbreak play a key role. For the Zimbabwe wave to be similar to India’s, there should have been similarities with the rest of the factors. In statistical parlance, all other factors should have been held constant. Including vaccination coverage, seroprevalence or wider restrictions, none of the factors is similar. It is purely a coincidence that the width and peak of one country are similar to those of another. Without proving this, the authors cannot go ahead with forecasting,” he said.
“The problem in identifying waves in India is that most States do not have robust surveillance systems. There is no clear way of understanding the expected number of cases of a disease. Therefore, it is difficult to explain the excess cases that occur,” Dr. Babu pointed out.
Concurring with the two experts, Prof. Athreya said that predictions solely based on statistical models should be read with caution. “It is vital to monitor cases closely, look for early warning signs, do extensive sequencing on test-positive cases and implement randomised sentinel based testing on sero-survey protocol,” he added.
